Content uploaded by Julien Kimmig
Author content
All content in this area was uploaded by Julien Kimmig on Jan 30, 2022
Content may be subject to copyright.
1
Biological Journal of the Linnean Society, 2022, XX, 1–18. With 6 figures.
Habitat and developmental constraints drove 330 million
years of horseshoe crab evolution
RUSSELL D. C. BICKNELL1,*,, JULIEN KIMMIG2,3,, GRAHAM E. BUDD4,
DAVID A. LEGG5, KENNETH S. BADER6, CAROLIN HAUG7,8, DORKAS KAISER9,10,,
LUKÁŠ LAIBL11,, JESSICA N. TASHMAN12, and NICOLÁS E. CAMPIONE1,
1Palaeoscience Research Centre, School of Environmental and Rural Science, University of New England,
Armidale, 2351 New South Wales, Australia
2Department of Geosciences, The Pennsylvania State University, University Park, PA 16802, USA
3Earth and Environmental Systems Institute, The Pennsylvania State University, University Park, PA
16802, USA
4Department of Earth Sciences, Palaeobiology Programme, Uppsala University, Villavägen 16, Uppsala,
SE 75236, Sweden
5Faculty of Science and Engineering, University of Manchester, Manchester, UK
6Jackson School of Geosciences, University of Texas, Austin, TX 78712, USA
7Ludwig-Maximilians-Universität München (LMU Munich), Biocenter, Großhaderner Str. 2, 82152
Planegg-Martinsried, Germany
8GeoBio-Center at LMU, Richard-Wagner-Str. 10, 80333 Munich, Germany
9Western Philippine University, Puerto Princesa City, 5300, Palawan, Philippines
10Katala Foundation Inc., Puerto Princesa City, Palawan, Philippines
11Czech Academy of Sciences, Institute of Geology, Rozvojová 269, 165 00 Prague 6, Czech Republic
12Department of Geology, Kent State University, 221 McGilvrey Hall, Kent, OH 44242, USA
Received 9 November 2021; revised 3 December 2021; accepted for publication 3 December 2021
Records of evolutionary stasis over time are central to uncovering large-scale evolutionary modes, whether by
long-term gradual change or via enduring stability punctuated by rapid shifts. The key to this discussion is to
identify and examine groups with long fossil records that, ideally, extend to the present day. One group often
regarded as the quintessential example of stasis is Xiphosurida, the horseshoe crabs. However, when, how and,
particularly, why stasis arose in xiphosurids remain fundamental, but complex, questions. Here, we explore
the protracted history of fossil and living xiphosurids and demonstrate two levels of evolutionary stability:
developmental stasis since at least the Pennsylvanian and shape stasis since the Late Jurassic. Furthermore,
shape and diversity are punctuated by two high-disparity episodes during the Carboniferous and Triassic –
transitions that coincide with forays into habitation of marginal environments. In an exception to these general
patterns, body size increased gradually over this period and, thus, cannot be described under the same, often-
touted, static models of evolution. Therefore, we demonstrate that evolutionary stasis can be modular and
fixed within the same group at different periods and in different biological traits, while other traits experience
altogether different evolutionary modes. This mosaic in the tempo and mode of evolution is not unique to
Xiphosurida but likely reflects variable mechanisms acting on biological traits, for example transitions in life
modes, niche occupation and major evolutionary radiations.
ADDITIONAL KEYWORDS: development – evolution – morphology – push of the past – stasis – Xiphosurida.
© The Author(s) 2022. Published by Oxford University Press on behalf of The Linnean Society of London.
All rights reserved. For permissions, please e-mail: journals.permissions@oup.com
applyparastyle “g//caption/p[1]” parastyle “FigCapt”
*Corresponding author. E-mail: rbickne2@une.edu.au
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
2 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
INTRODUCTION
Why are some species short-lived in the tree of life, rapidly
evolving into another, while others remain seemingly
unchanged for millions of years? Variation in the rates
of evolutionary change pervades the fossil record and is
fundamental to many modern evolutionary concepts,
including adaptive radiation, adaptive landscapes and
equilibria in general (Simpson, 1953; Gould & Eldredge,
1977; Eldredge et al., 2005; Lieberman & Eldredge, 2014;
Voje et al., 2015). Central to understanding these patterns
is the concept of stasis, often defined as little or no
morphological change over time (Levinton, 1983; Eldredge
et al., 2005). However, establishing why stasis might
become established in a group relies on first recognizing
when and how such patterns emerged, ideally by studying
living groups with extensive fossil records.
True horseshoe crabs (Xiphosurida) have long been
considered as ideal subjects to study stasis, as some fossil
species are notably similar to modern forms (Barthel,
1974; Bergström, 1975; Kin & Błażejowski, 2014;
Bicknell & Pates, 2020; Bicknell et al., 2021a), while
highly variable, ‘bizarre’ forms stand as clear evidence
against this mode of evolution across the Phanerozoic
(Eller, 1938; Eldredge, 1976; Selden & Siveter, 1987;
Haug et al., 2012; Haug & Rötzer, 2018b; Bicknell, 2019;
Bicknell & Pates, 2020; Haug & Haug, 2020; Bicknell
& Shcherbakov, 2021). Their long evolutionary history
and rare occurrences of exceptionally well-preserved
and abundant body fossils allow multiple facets of stasis
to be explored, including shape, size and development.
To date, these aspects of xiphosurid evolution remain
under-explored and as-of-yet not considered within a
single framework. Indeed, considering size, shape and
development in isolation is likely insufficient to justify
stasis in xiphosurid evolution (Eldredge et al., 2005).
Here, we present the first comprehensive empirical
evaluation of the growth, diversity, disparity and
body size in xiphosurids through time and seek to
uncover the more general patterns of their body-plan
evolution. We examine the long-running hypothesis
that stasis is the primary mode of xiphosurid evolution
(Barthel, 1974; Fisher, 1984; Kin & Błażejowski, 2014).
To do so, we test the null hypothesis that stasis has
been consistent within the group as far back as the
Carboniferous against the alternative hypothesis that
stasis arose after the end-Triassic, associated with the
extinction of abnormal xiphosurid morphologies. Our
study combines four datasets gathered from fossil and
extant xiphosurids: (1) geometric morphometric data to
describe shape and disparity, (2) linear measurement
data on body size, (3) generic diversity of the four major
families and (4) intraspecific linear measurement data
to reconstruct developmental allometries. Combined,
these datasets reveal the mosaic nature of stasis in
horseshoe crab evolution.
MATERIAL AND METHODS
The geometric morphometric data were collected from
photos of xiphosurid specimens with ages spanning the
Upper Mississippian (~331 Ma) to Recent. An analysis
using landmarks and semilandmarks of 178 specimens
was conducted following methods presented in earlier
works (Bicknell & Pates, 2019; Bicknell et al., 2019c;
Lustri et al., 2021). A total of 44 species, across 21
genera of Austrolimulidae, Belinuridae, Limulidae and
Paleolimulidae, were assessed. Generic assignments
primarily follow Bicknell & Pates (2020) and Lamsdell
(2020). Landmarking and semilandmarking were
conducted using the Thin-Plate Spline (TPS) suite
(Rohlf, 2015). A TPS file was constructed using
tpsUtil64 (v.1.7). The TPS file was imported into
tpsDig2 (v.2.26) and used to place three landmarks
on the prosoma, one landmark on the thoracetron
and 40 semilandmarks along the right prosomal
margin (Supporting Information, Fig. S1A; Table S1).
Semilandmarks were placed at regular intervals in
a clockwise direction along the anterior-most section
of the prosoma, starting at landmark 1 and ending
at the prosoma–thoracetron joint, landmark 3. Points
were digitized as xy coordinates and populated the
TPS file with landmark and semilandmark data
(Data S1). When the right side was poorly preserved,
the left side was digitized and then mirrored. The
TPS file was imported into R (v.4.0.0). The ‘geomorph’
package (v.3.2.1) (Adams et al., 2020) used the TPS
file to conduct the Procrustes superimposition,
optimize semilandmark positions using their bending
energies, and then ordinate Procrustes coordinates
through a principal component analysis (PCA; Table
S2; Supplementary Code 1). The results of the PCA
were output into Data S3, a meta-dataset including
temporal ranges for specimens, relevant references
for ages, proposed palaeoenvironment and formations
from which specimens were sampled. Procrustes
variances were calculated from the Procrustes
coordinates using ‘geomorph’ to estimate disparity in
each epoch division. The first two PCs were primarily
considered for exploring shape information as they
explained 62.4% of the data variation. PCs 3 and 4
were also examined (Figs S3–S5) but showed no other
informative patterns, so they were not examined at
length. Note that since disparity was calculated from
the coordinates, it explored the total variance in the
dataset, which is equivalent to using all PC axes.
Whether xiphosurids such as austrolimulids
and belinurids inhabited exclusively freshwater
conditions (sensu Lamsdell, 2016) is worth
consideration. Taxa such as Austrolimulus fletcheri
Riek, 1955, Dubbolimulus peetae Pickett, 1984 and
Franconiolimulus pochankei Bicknell et al., 2021b likely
inhabited completely freshwater palaeoenvironments
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
UNCOVERING PATTERNS OF HORSESHOE CRAB EVOLUTION 3
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
(see Bicknell et al., 2021b; Bicknell and Smith, 2021).
Conversely, the conditions inhabited by Alanops
magnificus Racheboeuf et al., 2002, Euproops danae
and Paleolimulus signatus are now considered to
represent more marginal conditions than freshwater
(Clements et al., 2019; Leibach et al., 2021). As such,
we conservatively propose marginal marine life modes
for these forms as opposed to supporting a completely
freshwater palaeoenvironment.
Specimen occurrences were binned by their
associated epoch as eras were too coarse and ages
were too fine to produce informative disparity patterns
(Supporting Information, Fig. S2; Tables S5 and S6).
Nevertheless, the overall disparity trajectory at the
age level was consistent with the coarser epoch-level
binning, supporting use of the latter. The disparity
at each time-bin was calculated as ‘sampled-in-bin’,
whereby disparity is measured relative to the mean at
that time. The disparity was also calculated using the
grand mean approach, whereby it is measured relative
to the mean of the entire sample from all time-bins
(Tables S4 and S6; Data S4; Figure S2). Both disparity
calculation approaches revealed comparable results;
however, preference was given to the former approach
to avoid biasing the calculation by the exceedingly
well-sampled Pennsylvanian. Pair-wise comparisons of
disparity values for ‘sampled-in-bin’ and grand mean
approaches for epoch and age datasets were calculated
(Tables S3–S6) and tested the following hypotheses:
H0:Disp: Disparity values are the same, indicating
a similar level of morphological variation
between epochs, or
HA:Disp: Disparity values are statistically different,
indicating distinct levels of morphological
variation between epochs.
A non-parametric Procrustes analysis of variance
(ANOVA) was run on the Procrustes coordinates
using ‘geomorph’ to examine differences in overall
morphology between these epochs. Epochs were
treated as discrete classification variables. Pair-wise
comparisons of epoch pairs were calculated to test the
following hypotheses:
H0:Morph: Mean shapes are the same, indicating a similar
average morphology between epochs, or
HA:Morph: Mean shapes are statistically different,
indicating distinct average morphologies
between epochs.
As framed here, the null hypothesis conforms
to stasis predictions of ‘little or no net accrued …
morphological change’ (Eldredge et al., 2005), whereas
the alternative would reject stasis as the primary
mode of shape evolution.
Prosomal and thoracetron lengths were collated
to represent body size and were measured in
ImageJ (v.1.52a) (Schneider et al., 2012). All linear
measurements were natural-log normalized, as there
is up to two orders-of-magnitude size difference across
the fossil and extant forms (Supporting Information,
Data S3).
A genus-level observation dataset was collated for
xiphosurids to explore diversity patterns spanning
the Mississippian to Recent (Supporting Information,
Data S5, S6). Age data for genera were derived
from ages presented in recent articles (Bicknell
& Pates, 2020; Bicknell et al., 2021b; Bicknell &
Shcherbakov, 2021) and the age data presented in
Data S3. Unfortunately, horseshoe crab genera are
largely represented by single occurrences and the
actual number of all xiphosurid specimens in each
time-bin, which is the sampling proxy needed to
undertake subsampling rarefaction analyses, is
currently unknown. As such, we opted to consider
only the presence or absence of genera in each epoch.
This approach means that diversity patterns will, in
part, be a function of sampling. However, the diversity
data only provide context for the disparity and size
time-series, which are less affected by sampling
biases, so overlooking sampling biases is unlikely to
affect the overall goals of this study. We focused on
the generic level, as the species-level designation of
xiphosurids is currently in flux and more likely to be
driven by the taxonomic philosophies of individual
authors (Bicknell & Pates, 2020; Haug & Haug, 2020;
Lamsdell, 2020).
The final dataset explores growth allometry and
consists of linear measurements from the prosomal
and thoracetronic sections of five fossil and one extant
species (Table 1; Supporting Information, Fig. S1B,
C). The fossil species represent the exceptionally rare
cases where >15 specimens are preserved in the same
deposit (see Fig. 1 for examples of each species). The
datasets were collated from either digital measurement
in ImageJ (Schneider et al., 2012) or callipers (Table
1; Data S7). These datasets were imported into R
and natural-log normalized. Differences in prosoma
and thoracetron growth trajectories were analysed
using the ‘smatr’ (v.3.4-8.) package (Warton et al.,
2012) and standardized major axis (SMA) line-fitting,
a model II regression that minimizes the residual
variances in both x and y axes without assuming an
independent–dependent relationship. Therefore, it is
the most appropriate means of reconstructing linear
relationships between the two morphometric variables
and inferring their allometric trajectories (Warton
et al., 2006).
The statistical similarities (or differences) in the
slopes (rates) and elevations (relative magnitudes) of
the prosoma and thoracetron datasets were tested as
follows. Slopes were tested through a likelihood ratio
test, in which:
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
4 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
Table 1. Meta-data on the species in the allometry dataset.
Species Family Formation, locality Age Collections used Data collector Means of
collection
Sample
size (N)
Euproops danae (Meek
& Worthen, 1865)
Belinuridae Mazon Creek
Lagerstätte,
Carbondale
Formation,
Illinois, USA
Moscovian,
Pennsylvanian
YMP IP, USNM Bicknell ImageJ 153
Euproops danae Belinuridae Mercer Shale,
Pennsylvania,
USA
Atokan, Pennsylvanian USNM Tashman,
originally from
Tashman et al.
(2019)
ImageJ 29
Euproops sp. Belinuridae Osnabrück
Formation,
Germany
Kasimovian,
Pennsylvanian
MAS Pal. Bicknell, originally
from Haug et al.
(2012)
ImageJ 18
Mesolimulus walchi
(Desmarest, 1822)
Limulidae Solnhofen
Limestone,
Germany
Late Kimmeridgian –
early Tithonian, Late
Jurassic
CM, JME SOS, MCZ,
MNHN, NM, SMNS,
SNSB-BSPG,
USNM, YPM IP
Bicknell ImageJ 104
Paleolimulus signatus
(Beecher, 1904)
Paleolimulidae Pony Creek Shale
Konservat-
Lagerstätte, Wood
Siding Formation,
Kansas, USA
Kasimovian,
Pennsylvanian
KUMIP, USNM, YMP
IP
Kimmig Callipers 49
Prolimulus woodwardi
Frič, 1899;
Belinuridae Kladno Formation,
Kladno, Czech
Republic
Moscovian,
Pennsylvanian
MCZ, NHMUK, NM,
MB.A
Bicknell ImageJ 16
Tachypleus tridentatus
(Leach, 1819)
Limulidae Bernardo Marcelo
Beach, Puerto
Princesa City,
Palawan,
Philippines
Recent N/A Kaiser, originally
from Kaiser &
Schoppe (2018)
Callipers 672
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
UNCOVERING PATTERNS OF HORSESHOE CRAB EVOLUTION 5
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
H0:Rate: Slope values are the same, indicating consistent
allometric trajectories or growth rates, or
HA:Rate: Slope values are different, indicating differential
allometric trajectories or growth rates.
If H0:Rate is supported, a common slope can be
calculated to test elevation differences (Warton et al.,
2006).
Elevations were tested using a t-test, in which:
H0:Mag: Elevation values are the same, indicating
consistent relative magnitudes between lengths
and widths, or
HA:Mag: Elevation values are different, indicating
differences in the relative length to width
magnitudes.
Figure 1. Examples of species considered for the allometry dataset. A, B, the belinurid Euproops danae. A, specimen from the
Mercer Shale, Pennsylvania, USA (Pennsylvanian, Atokan). USNM PAL 697642. B, specimen from the Mazon Creek Lagerstätte,
Carbondale Formation, Illinois, USA (Pennsylvanian, Moscovian). YPM IP 168016. C, Euproops sp. from the Osnabrück Formation,
Germany (Pennsylvanian, Kasimovian). MAS Pal. 484. D, the paleolimulid Paleolimulus signatus from the Pony Creek Shale
Konservat-Lagerstätte, Wood Siding Formation, Kansas, USA (Pennsylvanian, Kasimovian). USNM PAL 465528. E, the limulid
Mesolimulus walchi from the Solnhofen Limestone, Germany (Late Jurassic, late Kimmeridgian – early Tithonian). SNSB-BSPG
1894.I.1. F, the belinurid Prolimulus woodwardi from the Kladno Formation, Czech Republic (Pennsylvanian, Moscovian). NHMUK
In 18588. G, the limulid Tachypleus tridentatus, male. YPM IZ 55603. Photo credit: A, B, D, E, G, Russell Bicknell; C, Carolin Haug;
E, Lucie Goodayle and the NHMUK. C, converted to greyscale. Scale bars: A, 3 mm; B, C, E, 10 mm; D, F, 5 mm; G, 80 mm.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
6 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
INSTITUTIONAL ACRONYMS
AM F: Australian Museum, Sydney, NSW, Australia.
CM: Carnegie Museum of Natural History,
Pittsburgh, PA, USA. JME SOS: Jura-Museum,
Eichstätt, Germany. KUMIP: Division of Invertebrate
Paleontology, Biodiversity Institute, University of
Kansas, Kansas City, KS, USA. MAS Pal.: Museum am
Schölerberg, Osnabrück, Germany. MB.A.: Museum
für Naturkunde, Leibniz-Institut für Evolutions-
und Biodiversitätsforschung, Berlin, Germany. MCZ:
Museum of Comparative Zoology, Harvard University,
Cambridge, MA, USA. MNHN: Museum National
d’Histoire Naturelle of Paris, Paris, France. NHMUK:
Natural History Museum, London, UK. NM: National
Museum, Paleozoic Invertebrate collection, Prague,
Czech Republic. SMNS: State Museum of Natural
History Stuttgart, Stuttgart, Germany. SNSB-BSPG:
Staatliche Naturwissenschaftliche Sammlungen
Bayerns–Bayerische Staatssammlung für
Paläontologie und Geologie, Munich, Germany. USNM:
Smithsonian National Museum of Natural History,
Washington, DC, USA. UTGD: Geology Department,
University of Tasmania, Hobbart, Tas, Australia. YPM
IP: Division of Invertebrate Paleontology in the Yale
Peabody Museum, New Haven, CT, USA. YPM IZ:
Division of Invertebrate Zoology in the Yale Peabody
Museum, New Haven, CT, USA.
RESULTS
Generic xiphosurid diversity through time shows
three main peaks (Fig. 2). The first is during the
Middle and Upper Pennsylvanian, driven by high
belinurid diversity that decreases into the Cisuralian.
The second is the Lower to Middle Triassic, with large
diversifications in limulid and austrolimulid generic
taxa. Finally, there was a large increase in limulid
generic diversity during the Late Cretaceous. The
Cisuralian is the only time-bin when more than two
xiphosurid families are co-current.
PC1 represents 39.6% of the shape variance and
describes a continuum from atrophied to hypertrophied
genal spines (Fig. 3). The hypertrophied genial spines
of primarily Triassic austrolimulids and belinurids
with genal spines (e.g. Euproops; Fig. 1A–C) dominate
positive PC1 space, whereas limulids, paleolimulids
and belinurids lacking genal spines (e.g. Prolimulus;
Fig. 1F) and species with less pronounced or vestigial
Figure 2. Diversity of xiphosurids from the Mississippian through to today. Two main peaks are observed: during the
Middle to Upper Pennsylvanian and the Lower to Middle Triassic, transitions associated with the rise of non-marine forms.
In these time periods, at least two xiphosurid families are noted. A third peak is associated with high limulid diversity
during the Cretaceous and the evolution of two extant genera.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
UNCOVERING PATTERNS OF HORSESHOE CRAB EVOLUTION 7
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
genal spines (e.g. 4A) dominate negative PC1 space.
PC2 represents 22.8% of the shape variance and
describes the relative proportion of width to length,
with positive values denoting the typical limulid
morphology (=narrower and longer), while negative
values denote wider and shorter prosomal sections.
Furthermore, PC2 relates to the degree of posterior
genal spine elongation and relative position of
landmark 3 to the tip of the spine (Fig. 3).
The distribution of taxa through time shows an
extensive occupation of morphospace during the
Pennsylvanian and the Triassic and a much more limited
spread leading up to the Recent (Fig. 4B). These increases
in morphospace occupation coincide with the increased
generic diversity of belinurids and austrolimulids during
the Pennsylvanian and the Triassic, respectively (Fig.
4B). As such, the high disparity is linked to explorations
of positive and negative PC1 and PC2 morphospaces,
respectively (Supporting Information, Table S3). These
peaks in disparity are mirrored by the results of the
Procrustes ANOVA, which illustrate significant shifts
in mean shape from the Palaeozoic/early Mesozoic to
the younger time-bins. Specifically, Middle and Late
Pennsylvanian and Early Triassic mean xiphosurid
shape is significantly different from the xiphosurids in
younger intervals (Table S7). Following this shift, there
is no statistically significant difference in xiphosurid
shape from the Middle Triassic to the Recent (Table S7).
Results of the pair-wise, in-bin disparity analyses
show strong statistical support (P < 0.05) for
different disparity regimes from the Middle and Late
Pennsylvanian and Middle Triassic, and the Late
Jurassic and Recent forms (Supporting Information,
Table S3). These values are also reflected in analyses of
the grand mean dataset (Table S4). Such results reject
null hypotheses that stasis was the dominant mode
of evolution across these time periods. Furthermore,
a transition to statistically insignificant disparity
values for geologically younger pairs demonstrates
that disparity was markedly lower at, or after, the Late
Jurassic, a possible transition to morphological stasis.
The growth allometries show overall increases in
interspecific size (represented by the different elevations)
from the smallest specimens (belinurids) to the largest
Figure 3. Principal component analysis plot of Xiphosurida morphospace showing PC1 and PC2. Austrolimulids and
belinurids dominate positive PC1 space. Limulids and paleolimulids are located in negative PC1 space.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
8 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
specimens (paleolimulids and limulids) and increases
in intraspecific size (‘regression’ line) from juvenile
to adult individuals (Figs 5A, B, 6). Standardized
major axis lines show that the sampled xiphosurids
cannot be differentiated from isometry (i.e. slope ≈ 1;
Table 2). Furthermore, the slopes for the prosomal
sections are statistically indistinguishable (Table 3).
Significant differences were noted in the slope of the
thoracetron sections between Mesolimulus walchi and
both Paleolimulus signatus and Tachypleus tridentatus.
Overall these results support H0 for the slope, suggesting
consistent growth rates of prosomal and thoracetronic
Figure 4. Distribution of xiphosurids at the generic level and through time within PC1 and PC2. A, generic distribution in
PC space. B, temporal distribution in PC space.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
UNCOVERING PATTERNS OF HORSESHOE CRAB EVOLUTION 9
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
proportions in the six sampled taxa. By contrast,
widespread differences were recovered when considering
elevations, with overall support for HA (Table 4). Except
for the prosomal comparisons between Euproops danae
and Euproops sp. and between Paleolimulus signatus
and Prolimulus woodwardi, elevations of prosomal linear
models are significantly different (Table 4). For instance,
Euproops sp. have consistently and significantly wider
prosomal sections than all other sampled xiphosurids,
whereas Prolimulus and Paleolimulus have the relatively
narrowest prosomal sections (Fig. 5C, D). Overall, elevations
are significantly below 0 (Table 2), indicating that widths
are consistently larger in magnitude than lengths. All
thoracetronic elevations are significantly below 0, but,
unlike the prosoma, the thoracetron exhibits less overall
residual variation and follows a much more constrained
relationship (Fig. 5).
DISCUSSION
StaSiS defined
Due to the variable means by which stasis can be
defined, clarity on the definition is required. We
Figure 5. Natural logged length and width measurements of taxa studied for the allometric analyses. A, Prosomal
allometry for all taxa. B, Thoracetron allometry for all taxa. C, prosomal allometry, excluding Tachypleus tridentatus data.
D, thoracetronic allometry, excluding T. tridentatus data.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
10 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
Figure 6. Summary of xiphosurid evolution from the Upper Mississippian through to today. The temporal ranges and
proposed environmental habitats for the four assessed families are included. General trends towards modern day show
stasis in development, a shift towards stasis in shape and an overall increase in size. Principal component data illustrate
the two major pulses of disparity at the Pennsylvanian and Middle Triassic, associated with the presence of marginal
marine taxa. After the Jurassic, morphospace is highly constrained. The disparity curve follows this pattern closely, with
low disparity from the Late Jurassic to Recent. Size data show a prominent transition towards larger forms in the Late
Mesozoic and into the Cenozoic, with a slight increase in the Triassic, reflecting abnormally large austrolimulid forms.
Data from linear regressions illustrate a distinct conserved pattern of development (slope), with a varying record of length
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
UNCOVERING PATTERNS OF HORSESHOE CRAB EVOLUTION 11
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
largely follow that proposed by Eldredge et al. (2005:
133): stasis is inferred when there is ‘little or no net
accrued species-wide morphological change during
a species-lineage’s existence up to millions of years’.
However, the fossil record of xiphosurids does not offer
the sampling or temporal resolution needed to address
lineage-level dynamics. Therefore, we adapt the
Eldredge et al. framework to the broader clade level
and require two predictions to support stasis. Across
static periods, we should (1) recover non-significant
changes in the average morphology and (2) find that
both periods have low disparity. The first expectation
is consistent with the ‘little to no change’ argument
typically associated with stasis. However, because we
do not have the temporal resolution to explore lineage-
level dynamics, the low disparity between compared
intervals will demonstrate that, even within a time-
bin, shape did not deviate markedly from the mean,
thus supporting stasis within the interval. Effectively,
the disparity in a given time-bin may serve as a general
proxy for finer-scale morphological patterns therein.
In this study, developmental trajectories relate to
the overall prosomal and thoracetronic growth rates
(slopes) and relative magnitudes (intercepts) within
a species as it grew (Table 1; Mirth et al., 2016). To
support developmental stasis, we expect allometric
trajectories to be statistically consistent across
time-bins.
implicationS for xiphoSurid evolution
The comprehensive datasets presented here, exploring
diversity, shape, development and body size, reveal that
the long-running assumption of stasis in horseshoe
crab evolution is substantially more nuanced than
previously thought, despite the pervasive references
to ‘living fossils’ (Barthel, 1974; Eldredge, 1976; Avise
et al., 1994; Kin & Błażejowski, 2014; Lidgard & Love,
2018). First, we illustrate that stasis in xiphosurid
evolution developed in at least two stages (Fig. 6). The
first stage was developmental, as revealed by near-
identical intraspecific allometric patterns in the rate of
change (slope) between the widths and lengths of the
prosoma and thoracetron (Table 2). Allometric slopes
are remarkably consistent across the sampled fossil
and living species, suggesting either a developmental
constraint or stabilizing selection on allometry that
extends as far back as the Pennsylvanian. Conversely,
the relative magnitudes between widths and lengths
varied interspecifically over the examined period,
especially during the Pennsylvanian, highlighting
notable variability between wide-short and narrow-
long segments. This variation is mirrored by the shape
data, which likewise indicate high disparity during
the Pennsylvanian and the Early–Middle Triassic as
a result of proportional variations in prosomal shape
(Riek, 1955; Anderson, 1994, 1996; Lamsdell, 2020;
Bicknell & Shcherbakov, 2021; Lustri et al., 2021). The
Table 2. Summary values for allometric trajectories. Slope values are about 1 and elevation values are all negative.
Variable Slope (m) 95% CI mElevation (b) 95% CI b R2
Euproops danae prosoma 0.972 0.923 to 1.02 −0.791 −0.947 to −0.636 0.931
Euproops danae thoracetron 1.07 1.03 to 1.11 −0.687 −0.794 to −0.580 0.939
Euproops sp. prosoma 1.02 0.859 to 1.22 −0.984 −1.54 to −0.432 0.938
Euproops sp. thoracetron 1.08 0.925 to 1.25 −0.718 −1.10 to −0.337 0.934
Mesolimulus walchi prosoma 0.978 0.947 to 1.01 −0.516 −0.655 to −0.378 0.973
Mesolimulus walchi thoracetron 1.02 0.970 to 1.06 −0.493 −0.675 to −0.310 0.951
Paleolimulus signatus prosoma 1.02 0.924 to 1.13 −0.522 −0.878 to −0.166 0.902
Paleolimulus signatus thoracetron 1.17 1.06 to 1.29 −0.809 −01.16 to −0.459 0.917
Prolimulus woodwardi prosoma 0.999 0.799 to 1.25 −0.416 −1.02 to 0.187 0.846
Prolimulus woodwardi thoracetron 1.03 0.782 to 1.36 −0.596 −1.34 to 0.147 0.762
Tachypleus tridentatus prosoma 0.990 0.984 to 0.996 −0.541 −0.565 to −0.516 0.994
Tachypleus tridentatus thoracetron 1.09 1.08 to 1.09 −0.885 −0.908 to −0.861 0.995
to width (intercept) for the two main exoskeletal sections. Specimens and data points are colour coded to match those
assigned to families, and matched to Figure 2. A, B, representatives of Belinuridae. A, Prolimulus woodwardi. NHMUK In
18588. B, Euproops danae. USNM PAL 697642. C, a representative of Paleolimulidae – Paleolimulus signatus. USNM PAL
465528. D, E, representatives of the extreme Austrolimulidae. D, Tasmaniolimulus patersoni Bicknell, 2019 (Jackey Shale,
Tasmania, Australia, Permian, Lopingian). UTGD 123979, holotype. E, Austrolimulus fletcheri (Hawkesbury Sandstone,
NSW, Australia, Triassic, Ladinian). AM F38274, holotype. F, G, representatives of the Limulidae. F, Mesolimulus walchi
(Nusplingen Plattenkalk, Germany, Late Jurassic, late Kimmeridgian – early Tithonian). SMNS 70204. G, Tachypleus
tridentatus (female, YPM IZ 55576). Scale bars: A–D, 5 mm; E, F, 20 mm; G, 80 mm. Image credit: A, Lucie Goodayle and the
NHM; B–D, G, Russell Bicknell; E, Josh White; H, Guenter Schweigert.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
12 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
Table 4. Results of the SMA analyses considering the differences in elevation values for pairs of species. Statistically significantly different combinations are in
bold type.
Prosomal elevation results
Thoracetron
elevation
results
Euproops danae Euproops sp. Mesolimulus
walchi
Paleolimulus
signatus
Prolimulus
woodwardi
Tachypleus
tridentatus
Euproops danae Test stat: 1.20
P value: 0.273
Test stat: 176
P value: <2.22e−16
Test stat: 396
P value: <2.22e−16
Test stat: 186
P value: <2.22e−16
Test stat: 475
P value: <2.22e−16
Euproops sp. Test stat: 0.0345
P value: 0.857
Test stat: 77.9
P value: <2.22e−16
Test stat: 130
P value: <2.22e−16
Test stat: 106
P value: <2.22e−16
Test stat: 123
P value: <2.22e−16
Mesolimulus
walchi
Test stat: 0.943
P value: 0.332
Test stat: 2.17
P value: 0.140
Test stat: 53.6
P value: 2.47e−13
Test stat: 15.9
P value: 6.66e−5
Test stat: 17.1
P value: 3.50e−5
Paleolimulus
signatus
Test stat: 97.7
P value: <2.22e−16
Test stat: 7.49
P value: 6.23 e−3
Test stat: 45.4
P value: 1.60e−11
Test stat: 0.887
P value: 0.346
Test stat: 75.5
P value: <2.22e−16
Prolimulus
woodwardi
Test stat: 0.0003
P value: 0.987
Test stat: 0.0137
P value: 0.906
Test stat: 1.73
P value: 0.188
Test stat: 10.3
P value: 1.33e−3
Test stat: 56.8
P value: 2.21e−7
Tachypleus
tridentatus
Test stat: 190
P value: <2.22e−16
Test stat: 16.2
P value: 5.69e−5
Test stat: 153
P value: <2.22e−16
Test stat: 492
P value: <2.22e−16
Test stat: 14.8
P value: 1.22e−4
Table 3. Results of the SMA analyses considering the differences in slope values for pairs of species. Statistically significantly different combinations are in bold
type.
Prosomal slope results
Thoracetron slope
results Euproops
danae
Euproops sp. Mesolimulus
walchi
Paleolimulus
signatus
Prolimulus
woodwardi
Tachypleus
tridentatus
Euproops danae Test stat: 0.362
P value: 0.546
Test stat: 0.0408
P value: 0.840
Test stat: 0.732
P value: 0.392
Test stat: 0.0598
P value: 0.807
Test stat: 0.472
P value: 0.492
Euproops sp. Test stat: 0.0169
P value: 0.896
Test stat: 0.296
P value: 0.586
Test stat: 0.00158
P value: 0.968
Test stat: 0.0335
P value: 0.855
Test stat: 0.170
P value: 0.680
Mesolimulus
walchi
Test stat: 2.62
P value: 0.106
Test stat: 0.584
P value: 0.444
Test stat: 0.636
P value: 0.425
Test stat: 0.0367
P value: 0.848
Test stat: 0.487
P value: 0.485
Paleolimulus
signatus
Test stat: 3.09
P value: 0.0787
Test stat: 0.914
P value: 0.339
Test stat: 6.53
P value: 0.0106
Test stat: 0.0312
P value: 0.860
Test stat: 0.364
P value: 0.546
Prolimulus
woodwardi
Test stat: 0.0641
P value: 0.800
Test stat: 0.0.835
P value: 0.773
Test stat: 0.0121
P value: 0.913
Test stat: 0.802
P value: 0.371
Test stat: 0.00715
P value: 0.933
Tachypleus
tridentatus
Test stat: 0.884
P value: 0.347
Test stat: 0.0138
P value: 0.906
Test stat: 7.96
P value: 0.00479
Test stat: 2.31
P value: 0.129
Test stat: 0.154
P value: 0.695
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
UNCOVERING PATTERNS OF HORSESHOE CRAB EVOLUTION 13
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
high disparity also reflects the extreme genal spine
shapes exhibited by xiphosurids during these periods
(i.e. PC1; Figs 1, 3, 6; Supporting Information, Tables
S3 and S4). After these episodes of high variability in
genal spine shape and overall body proportions, shape
disparity decreased, becoming constrained from the
Jurassic (in the Bathonian, at the latest) through to
today (Table S3). The fixation of low disparity during
the Jurassic can be attributed to the limulid body
shape, which originated during the Triassic, perhaps
filling morphological gaps left by the extinction of
paleolimulids and some austrolimulids (Bicknell et al.,
in press). In either case, the typical horseshoe crab body
shape was fixed in the Jurassic, representing the onset
of the second stage of stasis in this group. Furthermore,
these patterns conform to our predictions that clade-
level stasis is reflected in both consistent inter-bin and
invariant intra-bin morphologies.
To extend these observations further back into the
Palaeozoic, prosomal and thoracetronic growth rates
were likely determined during the Lower Ordovician,
at the origin of Xiphosura (Rudkin et al., 2008; Va n
Roy et al. 2010, 2015). Indeed, a morphotype broadly
comparable to modern limulids is known from the
Lower Ordovician Fezouata Biota (Van Roy et al., 2010).
This similarity suggests a deep origin of the conserved
xiphosurid growth rate, a possible developmental
constraint that has persisted through episodic periods
of unsuccessful forays into other areas of morphospace
(Bicknell & Pates, 2020). It is also possible that
xiphosurid growth rates exhibit a strong stabilizing
selection, although this would require variable rates
to be strongly non-adaptive and it is unclear why that
would be the case.
Disparate xiphosurid forms record a complex
evolutionary history of diversification, possibly
through habitation of marginal environments
and show evidence for differential heterochronic
patterns compared to extant horseshoe crab species
(Sekiguchi et al., 1982; Shuster Jr & Sekiguchi, 2003;
Lamsdell, 2016, 2021; Haug & Rötzer, 2018a). For
instance, some belinurids resemble juvenile forms
of modern xiphosurids with elongate tail spines,
expressed thoracetronic segmentation, hypertrophied
ophthalmic spines, and absence or reduced genal
spines (Racheboeuf et al., 2002; Haug et al., 2012;
Haug & Rötzer, 2018b; Haug & Haug, 2020; Lamsdell,
2021; Lustri et al., 2021). Conversely, austrolimulids,
the other primarily marginal-marine group, show
splayed and overdeveloped genal spines and often
reduced thoracetronic sections (Pickett, 1984;
Lamsdell, 2016; Lerner et al., 2017; Bicknell, 2019;
Bicknell et al., 2021b). The onset of hypertrophied
spines may reflect an adaptation to the unidirectional
fluid flow associated with marginal marine conditions
(Anderson, 1996; Bicknell & Pates, 2019). Alternatively,
the spines were employed in subaerial activity (Fisher,
1979). Regardless of the functional mechanism, these
structures were primary drivers of high xiphosurid
disparity and shape change in the examined dataset.
The extinction of austrolimulids and the occupation
of exclusively marine conditions records the marked
transition to shape stasis.
The progressive increase in body size through time
(Fig. 6) contrasts with the aforementioned stages
of stasis in xiphosurids. Body length (excluding the
telson) shifts from 3–45 mm in the Palaeozoic, to
17–24.5 cm during the Upper Jurassic and Palaeogene,
and finally to >1 m in modern forms. The observed
body size dynamics are consistent with a ‘driven’ trend
(McShea, 1994), whereby the minimum size limit
increases along with the average and upper limit.
Driven trends suggest an underlying selection pressure
towards larger body sizes, perhaps owing to predation
or intraspecific interactions (competition or sexual
selection) acting within the bounds of developmental
and shape constraints (Bicknell et al., 2019a).
on xiphoSurid inStarS
Continuous growth trajectories for fossil xiphosurids
contrast the well-delineated ontogenetic clusters (i.e.
instars) observed in Tachypleus tridentatus (compare
data points in Fig. 5A, B to 5C, D). The lack of clusters
may reflect the reality that fossil xiphosurids are
often represented by singular specimens, and even
when abundantly preserved, fossil samples are much
smaller than extant populations. Furthermore, early
ontogenetic stages of non-biomineralized arthropods
are seldom preserved, even when specimens are
abundant (Haug et al., 2012). Alternatively, the assessed
fossil populations represent similar developmental
stages. However, considering the extensive datasets
of select taxa from a single outcrop (Paleolimulus
signatus and Euproops danae) and their large size
ranges (Supporting Information, Data S7), it seems
unlikely that single-instar groups were sampled.
Another plausible option is that these deposits
record a ‘snapshot in time’. In this case, a subset of
the true population is preserved, therefore limiting
evidence for instars (Yang et al., 2021). Extreme, often
anoxic conditions are required for the exceptional
preservation of soft cuticles typical of xiphosurids
(Allison, 1988; Kemp & Trueman, 2003; Gaines et al.,
2012; Garson et al., 2012; Clements et al., 2019).
Only the larger individuals would be capable of
entering into, surviving within and being preserved
in such environments, unless they were transported
(e.g. Bath-Enright et al., 2021). Indeed, in the rare
circumstances where clustered horseshoe crabs (the
only evidence for breeding in fossil xiphosurids;
Shuster Jr et al., 2003) have been observed, there
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
14 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
are no smaller individuals (Ambrose & Romano,
1972; Fisher, 1979; Bicknell et al., 2019b; Lustri
et al., 2021). Finally, the Mazon Creek Lagerstätte
xiphosurids were probably subjected to a limited
degree of transport from their original habitat prior
to preservation (Clements et al., 2019; Bath-Enright
et al., 2021). This transport would have damaged
smaller individuals, but larger specimens could
have endured more transport before disarticulating,
increasing the abundance of the adult specimens and
limiting evidence of instars.
diSparity and the ‘puSh of the paSt’
High disparity early in the evolutionary history of a
group, such as that recovered here for xiphosurids, has
been documented across a range of clades throughout
the history of life and is particularly prevalent among
fossil groups (Hughes et al., 2013). This pattern can
reflect fortunate ecological opportunities at the origin
of a new group and/or developmental constraints that
inhibit the exploration of new morphospaces as time
passes (Hughes et al., 2013, and references therein).
In either case, the disparity should decrease as the
clade continues to exist, resulting in a ‘bottom heavy’
distribution of disparity over time. Hughes et al.
(2013) expressed that ecological opportunity and
developmental constraints are not mutually exclusive,
and, indeed, we support both as potential mechanisms
driving the protracted evolutionary history of
Xiphosurida. We identified ancient developmental
constraints in prosomal and thoracetronic intraspecific
growth rates over the duration of xiphosurid evolution
(Fig. 5). However, overall shape and proportions vary
interspecifically (Fig. 6), especially during earlier ages,
and disparity peaks correspond with forays into new
environments (Figs 2, 6). Interestingly, the ‘bottom-
heavy’ disparity pattern recovered here contrasts with
those known for terrestrial arthropods, such as insects,
which exhibit high disparity in younger deposits,
resulting in a ‘top-heavy’ distribution (Labandeira &
Sepkoski, 1993; Labandeira, 2005; Labandeira & Eble,
in press).
A ‘bottom-heavy’ distribution could also be
explained by the ‘push of the past’ model if diversity
and disparity are coupled (Budd & Mann, 2018; Lee
& Dorey, 2018). This model suggests that a clade
that survives for a substantial time is likely to show
early rates of diversification followed by acquiescence
(Budd & Mann, 2018). This time-dependent pattern
in diversification rates stems from the fact that a
successful clade (one that survives an extensive
period) is likely to have benefited from early
evolutionary success. This phenomenon is particularly
pervasive if only extant members are considered in
macroevolutionary analyses but can also affect large
extinct clades (Budd & Mann, 2018).
Xiphosurid diversity generally tracks disparity
across the observed period, suggesting initial support
for a push of the past model. However, there are two
points where disparity and diversity decouple: the
Cisuralian and the Late Cretaceous (Fig. 2). During
these epochs, diversity was high while disparity was
low. The Late Cretaceous diversity peak is particularly
inconsistent with the predictions of the push of the
past model and reflects the rise of modern genera
and the presence of select limulid genera that went
extinct at the end of the Cretaceous. These forms
are all morphologically similar and Limulus-like,
hence the low disparity. The Cisuralian decoupling is
harder to explain and could reflect either sampling or
a true record of low disparity. Indeed, despite its low
disparity, there is no statistical difference between
the Cisuralian disparity and any other sampled epoch
(Supporting Information, Table S3), which likely
stems from low statistical power. However, with the
extinction of Belinuridae by the middle Cisuralian
(Malz & Poschmann, 1993; Anderson & Selden, 1997)
and the rise of austrolimulid forms without extremely
hypertrophied features (Bicknell et al., 2020, 2021b),
Permian xiphosurids may record a transition to the
disparity patterns mirrored in the Jurassic to Recent
forms. However, we note that diversity and disparity
are influenced by the limited preservational potential
of xiphosurid exoskeletons and the requirement for
exceptional preservation to record these specimens.
Furthermore, deposits such as the Pennsylvanian
Coal Measures Formation are extensively sampled,
producing higher diversity than other time-bins (sensu
Whitaker & Kimmig, 2020). As such, true diversity
and disparity may have been higher in periods when
lower counts were observed.
Although diversity–disparity comparisons partially
support the push of the past model, we point to two
factors suggesting a non-stochastic view of horseshoe
crab evolution. The first is that members of the clade
driving the initial inferred disparity peak during the
Pennsylvanian (i.e. belinurids) were not long-lived
compared to other families, such as limulids (Fig. 6). As
a result, despite their initial ‘success’, they went extinct
and did not factor in the reduction in diversification rates
observed during the Mesozoic and Cenozoic. Similarly, the
second disparity peak during the Triassic could support
a push of the past scenario for limulids following the
end-Permian extinction event (as predicted by Budd &
Mann, 2018). Limulids originated and diversified at that
time and persist until today. However, unlike diversity,
the disparity was driven by austrolimulids, which went
extinct in the Early Jurassic (Bicknell et al., 2021b),
whereas limulid disparity was static throughout the
Mesozoic and Cenozoic. Accordingly, both disparity peaks
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
UNCOVERING PATTERNS OF HORSESHOE CRAB EVOLUTION 15
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
are not associated with the long-lived extant families, as
predicted by the push of the past model, but are driven
by shorter-lived extinct families. The second factor is that
both disparity peaks correlate with similar morphospace
occupations (Figs 4, 6) and incursions into non-marine
niches (DeWoody & Avise, 2000; Selden et al., 2019; Lustri
et al., 2021). The fact that two independent xiphosurid
families evolved similar morphologies (hypertrophied
spines) in association with similar environmental
conditions is strong support for convergent adaptations
(Brooks & McLennan, 2002) in lieu of the random factors
stipulated by the ‘push of the past’.
CONCLUSION
Horseshoe crabs are regarded as the epitome of
evolutionary stasis, yet this evolutionary mode
remained untested and uncontextualized. Through
the concerted exploration of multiple complementary
datasets, we discovered that the group maintains an
unprecedented record of developmental stasis since
at least the Pennsylvanian. Developmental stasis
relates to stable prosomal and thoracetronic growth
rates that persisted across the emergence of disparate
interspecific forms (shapes) during the Pennsylvanian
and the Triassic. Following the Triassic disparity peak,
xiphosurids converged upon a static form strikingly
similar to those from the early Palaeozoic. The
convergence on the ‘typical’ horseshoe crab form in the
Jurassic suggests strong stabilizing selection on shape,
while active trends towards larger body sizes may reflect
biotic factors acting within the bounds of developmental
constraints and shape selection. The Pennsylvanian
and Triassic disparity peaks represent convergent
evolutions of hypertrophied spines, probably in response
to occupations of marginal marine environments.
Ultimately, these forays were unsuccessful and saw to
the extinction of Palaeozoic belinurids and Mesozoic
austrolimulids. By contrast, marine body plans
established near the xiphosurid origin have persisted,
largely unchanged, to the present day.
ACKNOWLEDGEMENTS
We thank Albert Kollar, Andreas Hecker, Angelika
Leipner, Claire Mellish, Jana Bruthansova, Jessica
Cundiff, Jessica Utrup, Mark Florence, Mike Reich
and Susan Butts for access to collections and specimen
images. We thank Guenter Schweigert, Josh White and
Lucie Goodayle for images. Cortney Langley and Nick
Rose are thanked for assistance with data collection.
Wade Leibach, Nick Rose and Kristopher Super are
thanked for assistance in the field. Sabine Schoppe is
thanked for her help with collecting the Tachypleus
tridentatus dataset. Finally, we thank Jason Dunlop
and Serge Naugolnykh for their reviews that improved
the scope and focus of the manuscript. The authors
declare no competing interests.
funding
University of New England Postdoctoral Research Fellow
(to R.D.C.B.). Charles Schuchert and Carl O. Dunbar
Grants-in-Aid award (to R.D.C.B.). Betty Mayne Scientific
Research Fund (to R.D.C.B.). James R Welch Scholarship
(to R.D.C.B). Ernst Mayr Grant (to R.D.C.B). Institute of
Geology of the Czech Academy of Sciences RVO 67985831
(to L.L.). Center for Geosphere Dynamics grant UNCE/
SCI/006 (to L.L.). Czech Science Foundation project
20-23550Y (to L.L.). German Research Foundation
(DFG, HA 7066/3-1 to C.H.). Australian Research Council
DECRA (DE190101423 to N.E.C).
author contributionS
Conceptualization: R.D.C.B., J.K., D.A.L and N.E.C.
Methodology: R.D.C.B. and N.E.C. Datasets: R.D.C.B.,
J.K., D.A.L., K.S.B., C.H., D.K. and J.N.T. Investigation:
R.D.C.B., J.K., G.E.B. and N.E.C. Resources: R.D.C.B.
Writing – original draft: R.D.C.B., G.E.B., L.L.,
J.K. and N.E.C. Writing – review and editing: R.D.C.B.,
J.K., G.E.B., D.A.L., K.S.B., C.H., D.K., L.L., J.N.T. and
N.E.C. Visualization: R.D.C.B. and N.E.C. Funding
acquisition: R.D.C.B., C.H., L.L. and N.E.C.
The authors declare no competing interests.
DATA AVAILABILITY
All raw data, code and supplemental results are
available in the Supporting Information.
REFERENCES
Adams DC, Collyer ML, Kaliontzopoulou A. 2020.
Geomorph: Software for geometric morphometric analyses.
R package version 3.2.1. https://cran.r-project.org/
package=geomorph.
Allison PA. 1988. Konservat-Lagerstätten: cause and
classification. Paleobiology 14: 331–344.
Ambrose T, Romano M. 1972. New Upper Carboniferous
Chelicerata (Arthropoda) from Somerset, England.
Palaeontology 15: 569–578.
Anderson LI. 1994. Xiphosurans from the Westphalian D of
the Radstock Basin, Somerset Coalfield, the South Wales
Coalfield and Mazon Creek, Illinois. Proceedings of the
Geologists’ Association 105: 265–275.
Anderson LI. 1996. Taphonomy and taxonomy of Palaeozoic
Xiphosura. Unpublished PhD Thesis, University of Manchester.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
16 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
Anderson LI, Selden PA. 1997. Opisthosomal fusion and
phylogeny of Palaeozoic Xiphosura. Lethaia 30: 19–31.
Avise JC, Nelson WS, Sugita H. 1994. A speciational history
of ‘Living Fossils’: molecular evolutionary patterns in
horseshoe crabs. Evolution 48: 1986–2001.
Barthel KW. 1974. Limulus: a living fossil. Naturwissenschaften
61: 428–433.
Bath-Enright OG, Minter NJ, Summer EJ, Mángano MG,
Buatois LA. 2021. Flume experiments reveal flows in
the Burgess Shale can sample and transport organisms
across substantial distances. Communications Earth and
Environment 2: 104.
Beecher CE. 1904. Note on a New Permian Xiphosuran from
Kansas. American Journal of Science 18: 23–24.
Bergström J. 1975. Functional morphology and evolution of
xiphosurids. Fossils and Strata 4: 291–305.
Bicknell RDC. 2019. Xiphosurid from the Upper Permian
of Tasmania confirms Palaeozoic origin of Austrolimulidae.
Palaeontologia Electronica 22: 1–13.
Bicknell RDC, Błażejowski B, Wings O, Hitij T,
Botton ML. 2021a. Critical re-evaluation of Limulidae
reveals limited Limulus diversity. Papers in Palaeontology
7: 1525–1556.
Bicknell RDC, Brougham T, Charbonnier S, Sautereau F,
Hitij T, Campione NE. 2019a. On the appendicular anatomy
of the xiphosurid Tachypleus syriacus and the evolution of fossil
horseshoe crab appendages. The Science of Nature 106: 38.
Bicknell RDC, Hecker A, Heyng AM. 2021b. New horseshoe
crab fossil from Germany demonstrates post-Triassic
extinction of Austrolimulidae. Geological Magazine 158:
1461–1471.
Bicknell RDC, Naugolnykh SV, Brougham T. 2020. A
reappraisal of Paleozoic horseshoe crabs from Russia and
Ukraine. The Science of Nature 107: 46.
Bicknell RDC, Naugolnykh SV, McKenzie SC. in press. On
Paleolimulus from the Mazon Creek Konservat-Lagerstätte.
Comptes Rendus Palevol.
Bicknell RDC, Pates S. 2019. Xiphosurid from the
Tournaisian (Carboniferous) of Scotland confirms deep
origin of Limuloidea. Scientific Reports 9: 17102.
Bicknell RDC, Pates S. 2020. Pictorial atlas of fossil and
extant horseshoe crabs, with focus on Xiphosurida. Frontiers
in Earth Science 8: 98.
Bicknell RDC, Pates S, Botton ML. 2019b. Euproops danae
(Belinuridae) cluster confirms deep origin of gregarious
behaviour in xiphosurids. Arthropoda Selecta 28: 549–555.
Bicknell RDC, Shcherbakov DE. 2021. New austrolimulid
from Russia supports role of Early Triassic horseshoe crabs
as opportunistic taxa. PeerJ 9: e11709.
Bicknell RDC, Smith PM. 2021. The first fossil scorpion from
Australia. Alcheringa. DOI: 10.1080/03115518.03112021.01
983874.
Bicknell RDC, Žalohar J, Miklavc P, Celarc B, Križnar M,
Hitij T. 2019c. A new limulid genus from the Strelovec
Formation (Middle Triassic, Anisian) of northern Slovenia.
Geological Magazine 156: 2017–2030.
Brooks DR, McLennan DA. 2002. The nature of diversity:
an evolutionary voyage of discovery. Chicago: University of
Chicago Press.
Budd GE, Mann RP. 2018. History is written by the victors:
the effect of the push of the past on the fossil record. Evolution
72: 2276–2291.
Clements T, Purnell M, Gabbott S. 2019. The Mazon Creek
Lagerstätte: a diverse late Paleozoic ecosystem entombed
within siderite concretions. Journal of the Geological Society
176: 1–11.
Desmarest A-G. 1822. Les crustacés proprement dits. In:
Brongniart A, Desmarest A-G, eds. Histoire naturelle des
crustacés fossiles, sous les rapports zoologiques et geologiques.
Paris: F.-G. Levrault, 67–142.
DeWoody JA, Avise JC. 2000. Microsatellite variation in
marine, freshwater and anadromous fishes compared with
other animals. Journal of Fish Biology 56: 461–473.
Eldredge N. 1976. Differential evolutionary rates. Paleobiology
2: 174–177.
Eldredge N, Thompson JN, Brakefield PM, Gavrilets S,
Jablonski D, Jackson JBC, Lenski RE, Lieberman BS,
McPeek MA, Miller W III. 2005. The dynamics of
evolutionary stasis. Paleobiology 31: 133–145.
Eller ER. 1938. A review of the xiphosuran genus Belinurus
with the description of a new species, B. alleganyensis.
Annals of the Carnegie Museum 27: 129–150.
Fisher DC. 1979. Evidence for subaerial activity of Euproops
danae (Merostomata, Xiphosurida). In: Nitecki MH, ed.
Mazon Creek fossils. New York: Elsevier, 379–447.
Fisher DC. 1984. The Xiphosurida: archetypes of bradytely?
In: Eldredge N, Stanley SM, eds. Living fossils. New York:
Springer, 196–213.
Frič A. 1899. Preliminary note on Prolimulus woodwardi,
Fritsch, from the Permian Gaskohle at Nyřan, Bohemia.
Geological Magazine 6: 57–58.
Gaines RR, Hammarlund EU, Hou X, Qi C, Gabbott SE,
Zhao Y, Peng J, Canfield DE. 2012. Mechanism for
Burgess Shale–type preservation. Proceedings of the
National Academy of Sciences USA 109: 5180–5184.
Garson DE, Gaines RR, Droser ML, Liddell WD,
Sappenfield A. 2012. Dynamic palaeoredox and exceptional
preservation in the Cambrian Spence Shale of Utah. Lethaia
45: 164–177.
Gould SJ, Eldredge N. 1977. Punctuated equilibria: the
tempo and mode of evolution reconsidered. Paleobiology 3:
115–151.
Haug C, Haug JT. 2020. Untangling the Gordian knot–further
resolving the super-species complex of 300-million-year-old
xiphosurids by reconstructing their ontogeny. Development
Genes and Evolution 230: 13–26.
Haug C, Rötzer MAIN. 2018a. The ontogeny of Limulus
polyphemus (Xiphosura s. str., Euchelicerata) revised:
looking ‘under the skin’. Development Genes and Evolution
228: 49–61.
Haug C, Rötzer MAIN. 2018b. The ontogeny of the 300
million year old xiphosuran Euproops danae (Euchelicerata)
and implications for resolving the Euproops species complex.
Development Genes and Evolution 228: 63–74.
Haug C, Van Roy P, Leipner A, Funch P, Rudkin DM,
Schöllmann L, Haug JT. 2012. A holomorph approach to
xiphosuran evolution—a case study on the ontogeny of Euproops.
Development Genes and Evolution 222: 253–268.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
UNCOVERING PATTERNS OF HORSESHOE CRAB EVOLUTION 17
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
Hughes M, Gerber S, Wills MA. 2013. Clades reach
highest morphological disparity early in their evolution.
Proceedings of the National Academy of Sciences USA 110:
13875–13879.
Kaiser D, Schoppe S. 2018. Postembryonic development
of Tachypleus tridentatus (Merostomata: Xiphosura) in a
nursery habitat in the Philippines. Journal of Threatened
Taxa 10: 12916–12932.
Kemp RA, Trueman CN. 2003. Rare earth elements in
Solnhofen biogenic apatite: geochemical clues to the
palaeoenvironment. Sedimentary Geology 155: 109–127.
Kin A, Błażejowski B. 2014. The horseshoe crab of the
genus Limulus: living fossil or stabilomorph? PLoS ONE 9:
e108036.
Labandeira CC. 2005. The fossil record of insect extinction:
new approaches and future directions. American
Entomologist 51: 14–29.
Labandeira CC, Eble GJ. in press. The fossil record of insect
diversity and disparity. In: Anderson J, Thackeray F, Wyk BV,
Wit MD, eds. Gondwana alive: biodiversity and the evolving
biosphere. Johannesburg: Witwatersrand University Press.
Labandeira CC, Sepkoski JJ. 1993. Insect diversity in the
fossil record. Science 261: 310–315.
Lamsdell JC. 2016. Horseshoe crab phylogeny and
independent colonizations of fresh water: ecological invasion
as a driver for morphological innovation. Palaeontology 59:
181–194.
Lamsdell JC. 2020. The phylogeny and systematics of
Xiphosura. PeerJ 8: e10431.
Lamsdell JC. 2021. A new method for quantifying heterochrony
in evolutionary lineages. Paleobiology 47: 363–384.
Leach WE. 1819. Entomostraca. In: Levrault F, ed. Dictionaire
des Science Naturelles. Paris: Levrault and Schoell, 524–543.
Lee MSY, Dorey JB. 2018. Evolution: Dampening the
Cambrian explosion. Current Biology 28: R1353–R1355.
Leibach WW, Rose N, Bader K, Mohr LJ, Super K,
Kimmig J. 2021. Horseshoe crab trace fossils and associated
ichnofauna of the Pony Creek Shale Lagerstätte, Upper
Pennsylvanian, Kansas, USA. Ichnos 28: 34–45.
Lerner AJ, Lucas SG, Lockley M. 2017. First fossil
horseshoe crab (Xiphosurida) from the Triassic of North
America. Neues Jahrbuch für Geologie und Paläontologie-
Abhandlungen 286: 289–302.
Levinton JS. 1983. Stasis in progress: the empirical basis of
macroevolution. Annual Review of Ecology and Systematics
14: 103–137.
Lidgard S, Love AC. 2018. Rethinking living fossils.
BioScience 68: 760–770.
Lieberman BS, Eldredge N. 2014. What is punctuated
equilibrium? What is macroevolution? A response to Pennell
et al. Trends in Ecology and Evolution 29: 185–186.
Lustri L, Laibl L, Bicknell RDC. 2021. A revision of
Prolimulus woodwardi Fritsch, 1899 with comparison to
other paedomorphic belinurids. PeerJ 9: e10980.
Malz H, Poschmann M. 1993. Erste Süßwasser-Limuliden
(Arthropoda, Chelicerata) aus dem Rotliegenden der
Saar-Nahe-Senke. Osnabrücker naturwissenschafliche
Mitteilungen 19: 21–24.
McShea DW. 1994. Mechanisms of large-scale evolutionary
trends. Evolution 48: 1747–1763.
Meek FB, Worthen AH. 1865. Notice of some new types
of organic remains, from the Coal Measures of Illinois.
Proceedings of the Academy of Natural Sciences of
Philadelphia 17: 41–48.
Mirth CK, Frankino WA, Shingleton AW. 2016. Allometry
and size control: what can studies of body size regulation
teach us about the evolution of morphological scaling
relationships? Current Opinion in Insect Science 13: 93–98.
Pickett JW. 1984. A new freshwater limuloid from the middle
Triassic of New South Wales. Palaeontology 27: 609–621.
Racheboeuf PR, Vannier J, Anderson LI. 2002. A new
three-dimensionally preserved xiphosuran chelicerate
from the Montceau-les-Mines Lagerstätte (Carboniferous,
France). Palaeontology 45: 125–147.
Riek EF. 1955. A new xiphosuran from the Triassic sediments
at Brookvale, New South Wales. Records of the Australian
Museum 23: 281–282.
Rohlf FJ. 2015. The tps series of software. Hystrix 26: 9–12.
Rudkin DM, Young GA, Nowlan GS. 2008. The oldest
horseshoe crab: a new xiphosurid from Late Ordovician
Konservat-Lagerstätten deposits, Manitoba, Canada.
Palaeontology 51: 1–9.
Schneider CA, Rasband WS, Eliceiri KW. 2012. NIH Image
to ImageJ: 25 years of image analysis. Nature Methods 9:
671–675.
Sekiguchi K, Yamamichi Y, Costlow JD. 1982. Horseshoe
crab developmental studies I. Normal embryonic
development of Limulus polyphemus compared with
Tachypleus tridentatus. In: Bonaventura J, Bonaventura C,
Tesh S, eds. Physiology and Biology of Horseshoe crabs:
Studies on Normal and Eenvironmentally Stressed Animals.
New York, 53–73.
Selden PA, Simonetto L, Marsiglio G. 2019. An effaced
horseshoe crab (Arthropoda: Chelicerata: Xiphosura) from
the Upper Carboniferous of the Carnic Alps (Friuli, NE
Italy). Rivista Italiana di Paleontologia e Stratigrafia 125:
333–342.
Selden PA, Siveter DJ. 1987. The origin of the limuloids.
Lethaia 20: 383–392.
Shuster CN Jr, Barlow RB, Brockmann HJ. 2003. The
American horseshoe crab. Cambridge: Harvard University
Press.
Shuster CN Jr, Sekiguchi K. 2003. Growing up takes
about ten years and eighteen stages. In: Shuster CN, Jr,
Barlow RB, Brockmann HJ, eds. The American horseshoe
crab. Cambridge: Harvard University Press, 103–132.
Simpson GG. 1953. The major features of evolution. New York:
Columbia University Press.
Tashman JN, Feldmann RM, Schweitzer CE. 2019.
Morphological variation in the Pennsylvanian horseshoe
crab Euproops danae (Meek & Worthen, 1865) (Xiphosurida,
Euproopidae) from the lower Mercer Shale, Windber,
Pennsylvania, USA. Journal of Crustacean Biology 39:
396–406.
Van Roy P, Briggs DEG, Gaines RR. 2015. The Fezouata
fossils of Morocco; an extraordinary record of marine life in
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
18 R. D. C. BICKNELL ET AL.
© 2022 The Linnean Society of London, Biological Journal of the Linnean Society, 2022, XX, 1–18
the Early Ordovician. Journal of the Geological Society 172:
541–549.
Van Roy P, Orr PJ, Botting JP, Muir LA, Vinther J,
Lefebvre B, El Hariri K, Briggs DEG. 2010. Ordovician
faunas of Burgess Shale type. Nature 465: 215–218.
Voje KL, Holen ØH, Liow LH, Stenseth NC. 2015. The role
of biotic forces in driving macroevolution: beyond the Red
Queen. Proceedings of the Royal Society. Series B, Biological
Sciences 282: 20150186.
Warton DI, Duursma RA, Falster DS, Taskinen S. 2012.
smatr 3–an R package for estimation and inference about
allometric lines. Methods in Ecology and Evolution 3: 257–259.
Warton DI, Wright IJ, Falster DS, Westoby M. 2006.
Bivariate line-fitting methods for allometry. Biological
Reviews 81: 259–291.
Whitaker AF, Kimmig J. 2020. Anthropologically introduced
biases in natural history collections, with a case study on
the invertebrate paleontology collections from the middle
Cambrian Spence Shale Lagerstätte. Palaeontologia
Electronica 23: a58.
Yang X, Kimmig J, Zhai D, Liu Y, Kimmig SR, Peng S. 2021.
A juvenile-rich palaeocommunity of the lower Cambrian
Chengjiang biota sheds light on palaeo-boom or palaeo-bust
environments. Nature Ecology and Evolution 5: 1082–1090.
SUPPORTING INFORMATION
Additional Supporting Information may be found in the online version of this article at the publisher’s web-site:
Data S1: The TPS file of specimens analysed for the geometric morphometric dataset.
Data S2: CSV file for semilandmark sliding used in the geometric morphometric analysis.
Data S3: CSV file of PCA results. Proposed palaeoecology, familial, generic, stratigraphic, temporal and
measurement data are included. References for temporal constraints are also included.
Data S4: CVS file of disparity values and time-bins.
Data S5: CSV file for diversity counts.
Data S6: CSV file for combined diversity data.
Data S7: XLS file of all allometry datasets.
Supplemental Code 1: R code used for analyses.
Figure S1: Depiction of data gathered for both datasets. A, semilandmark trajectory (blue dotted line) and four
digitized landmarks. B, linear measurements taken for Mesolimulus walchi, Paleolimulus signatus, Prolimulus
woodwardi and Tachypleus tridentatus specimens. C, linear measurements taken for Euproops danae and
Euproops sp. Genal spines are not measured in these species as these are hypertrophied and are seldom preserved
on both prosomal sides. Numbers in A correspond to landmarks. Abbreviations: PL, prosomal length; PW, prosomal
width; TL, thoracetronic length; TW, thoracetronic width.
Figure S2: Additional disparity plots of epoch and age-level bins. Both the sampled-in-bin and grand mean
approaches are shown.
Figure S3: PCA plot of Xiphosurida morphospace showing PC1 and PC3 coded for family groups.
Figure S4: PCA plot of Xiphosurida morphospace showing PC2 and PC3 coded for family groups.
Figure S5: PCA plot of Xiphosurida morphospace showing PC3 and PC4 coded for family groups.
Table S1: Description of landmarks used for the geometric morphometric analysis. Figure S1A depicts these
landmarks.
Table S2: Proportion of variances explained by PC.
Table S3: P-values from disparity analyses using epoch time-bins and in-bin sampling. Bold values indicate
statistically significant differences in disparity for epoch pairs.
Table S4: P-values from disparity analyses using epoch time-bins and the grand mean. Bold values indicate
statistically significant differences in disparity for epoch pairs.
Table S5: P-values from disparity analyses using age time-bins and in-bin sampling. Bold values indicate
statistically significant differences in disparity for age pairs.
Table S6: P-values from disparity analyses using age time-bins and the grand mean. Bold values indicate
statistically significant differences in disparity for age pairs.
Table S7: P-values from pair-wise comaprisions of analysis of vairance (ANOVA) of aligned landmark and
semilandmark data. Bold values indicate statistically significant differences in means for epoch pairs, indicating
marked difference in morphology between epochs.
Downloaded from https://academic.oup.com/biolinnean/advance-article/doi/10.1093/biolinnean/blab173/6513746 by Pennsylvania State University user on 30 January 2022
